Most lay-people see photography as the cut-and-dried act of capturing the world “as it is”. Ironically, the simple and necessary action of framing a photograph instantly renders the result subjective (’un-photogenic’ subjects are typically pushed out-of-frame). This creative choice built into the camera’s design is part of the appeal of photography. However, we occasionally encounter entirely photogenic scenes that we wish to convey in their entirety. Luckily, there is a type of photograph that has no edges at all; the spherical panorama.
This spherical panorama captures the complete scene; from the zenith (above my head) to the complete horizon to the the nadir (under my feet). No photographic paraphernalia (camera, tripod, photographer) distracts the viewer from the scene. What kind of magical camera does this? Well, with the right know-how, software and technique any camera will do. This is because spherical panoramas are simply an amalgamation of many regular photographs. Here’s the kit I used to make my example panorama.
Tripod fitted with a Nodal Ninja 5 panoramic tripod head supporting a Canon 5D mark II with Canon 16 - 35mm f2.8L lens
To get started we need a sequence of regular photographs which cover the entire scene. Typically this involves taking several ‘rows’ of images, each containing a constant number of images (columns). The number of rows and columns required depends on the lens used but nevertheless the angle between rows should be consistent, as should that between each column (made easy by using a panoramic tripod head, e.g. Nodal Ninja 5). In addition, a zenith shot (directly up) and two (complimentary) ‘nadir’ shots (directly down) are required. Here’s the sequence of images I used to produce my sample panorama.
zenith
sky row
horizon row
ground row
nadir (taking two at opposites enables later removal of feet & shadows)The rest of the spherical panorama creation process is all in software. I plan to cover the details in a future screen-cast, but for now here’s a quick overview:
Ensure ‘consistent colour’ across shots by correcting white-balance discrepancies (tip: remember to lock white-balance, exposure, shutter-speed, focal length and focus while shooting). Also correct any vignetting (dark edges) and chromatic aberration (coloured fringing on detail). For these tasks I use Canon’s bundled DPP software (if Lightroom’s vignetting correction tools were better, I’d use it instead).
Import the corrected images into a stitching tool (after trying a few apps I finally ponied up for a copy of PTGui). By creating control-point pairs (i.e. locating similar features in different photographs) I give PTGui the necessary information to correctly align, warp, composite and blend the all images (except for the zenith and nadir) into a single ’stitched’ equirectilinear result.- In photoshop I work a little magic to combine the two original nadir shots (which both contain my legs and feet) into a single feet-free nadir image
- Using PTGui again I produce another equirectilinear image, this time containing only the warped ground (from the composited nadir) and sky (from the zenith shot)
Back in photoshop I combine both equirectilinear images into a single image (’patching’) and apply any necessary retouching (e.g. making horizons dead-level and apply colour-corrections) to produce the pièce de résistance: the final (70+ megapixel) image.
patching the warped nadir and zenith onto the stitched panorama in photoshop
At first glance this appears to be a regular photo, but in fact it covers 360° horizontally and 180° vertically (i.e. the entire scene). So how does it justify the ’spherical’ and ‘edge-less’ tags? Well, the most effective way to present such an image is to wrap the rectangular image around an imaginary sphere and give the viewer control of a virtual camera placed at the orb’s centre. Rotating the camera simulates ‘looking’ in the original scene. There are no limits on where the camera may look and no visible edges or seams.
the distortion in the equirectilinear image disappears when the image is wrapped around a sphere (inside which the viewer’s virtual ‘camera’ rotates)
Spherical panoramas really take on a new dimension when combined with software like the PangeaVR iPhone app I’ve mentioned before. Wherever I am I can show friends exactly what my favourite places are like by handing them the phone and saying “drag the image with your finger”. No other explanation, instruction or intervention is required for me to show friends overseas what the beach I learned to surf at is “really like”. Of course, the panoramas can also be viewed in a web browser. Below are links to a few panoramas I’ve created in the past few months from images taken on my Christmas holiday back home.
 |
 |
 |
 |
 |
 |
 |
 |
 |
Hopefully I’ll have more to add to my panoramas page soon.
PangeaVR is (amazingly) free, though I 
Download and extract 
Sharpness is the top of the list for most lens geeks. Not all lenses are created equal, especially considering the difference in the ’sharpness’ of the image they produce. A single lens can vary greatly between different focal length/aperture combinations and even varies (non-linearly) from centre to corner within a single shot!
Distortion is what causes real world straight lines to appear curved in your photographs. Zoom lenses in particular are prone to distorting ‘inwards’ (pincushion) at one extreme and ‘outwards’ (barrel) at the other. Distortion varies by focal length.
Falloff is the phenomenon whereby the corners of the frame appear ‘darker’ than the centre (i.e. the light ‘falls off’ from centre to edge). Though often used as an intentional effect, falloff is generally avoided if possible and varies with focal length, aperture and obviously, distance from centre.
There was much internal debate (which still rages) about this UI element which allows users to change/compare reviews, but the eventual decision was to filter reviews by system, lens and finally camera body. Note: Our lens tests are performed on a camera body (instead of an optical test bench) which has relevance upon the results.
